Crises as Opportunities for Growth: The Strategic Value of Business Group Affiliation - European ...
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Crises as Opportunities for Growth: The Strategic Value of Business Group Affiliation Finance Working Paper N° 781/2021 Ronald W. Masulis University of New South Wales, ABFER, FIRN, September 2021 FMA , National University of Singapore, and ECGI Peter K. Pham University of New South Wales and FIRN Jason Zein University of New South Wales and FIRN Alvin E. S. Ang The Hang Seng University of Hong Kong and University of New South Wales © Ronald W. Masulis, Peter K. Pham, Jason Zein and Alvin E. S. Ang 2021. All rights reserved. Short sections of text, not to exceed two paragraphs, may be quoted without explicit permission provided that full credit, including © notice, is given to the source. This paper can be downloaded without charge from: http://ssrn.com/abstract_id=2517810 www.ecgi.global/content/working-papers
ECGI Working Paper Series in Finance Crises as Opportunities for Growth: The Strategic Value of Business Group Affiliation Working Paper N° 781/2021 September 2021 Ronald W. Masulis Peter K. Pham Jason Zein Alvin E. S. Ang © Ronald W. Masulis, Peter K. Pham, Jason Zein and Alvin E. S. Ang 2021. All rights reserved. Short sections of text, not to exceed two paragraphs, may be quoted without explicit permission provided that full credit, including © notice, is given to the source.
Abstract We document a novel strategic motive for family business groups to utilize their internal capital markets (ICMs) during financial crises. We find that crisis-period group ICM activity is targeted toward exerting product market dominance over standalone rivals. Groups make significant post-crisis gains in market share that are concentrated among affiliates (and industry segments within affiliates) operat- ing in highly competitive product markets, where capturing such gains is difficult in normal times. These patterns are observed only in emerging markets, suggesting that ICMs enable groups to exploit crises to realize long-term competitive advan- tages only when rivals face chronic financing frictions. Keywords: Family Business Groups, Product-market competition, Financial crisis JEL Classifications: G01, G31, G32 Ronald W. Masulis Scientia Professor in Finance School of Banking and Finance University of New South Wales Business School Gate 2 High Street, Kensington Campus Sydney, NSW 2052, Australia phone: + 61 (2) 9385 5860 e-mail: ron.masulis@unsw.edu.au Peter K. Pham* Associate Professor School of Banking and Finance University of New South Wales Business School Gate 2 High Street, Kensington Campus UNSW Sydney, NSW 2052, Australia phone: +61 (2) 9385 5875 e-mail: peter.pham@unsw.edu.au Jason Zein Associate Professor School of Banking and Finance University of New South Wales Business School Gate 2 High Street, Kensington Campus UNSW Sydney, NSW 2052, Australia e-mail: j.zein@unsw.edu.au Alvin E. S. Ang Assistant Professor The Hang Seng University of Hong Kong Department of Economics and Finance, School of Business, Shatin, Hong Kong phone: +(852) 3963 5653 e-mail: alvinang@hsu.edu.hk *Corresponding Author
Crises as Opportunities for Growth: The Strategic Value of Business Group Affiliation Ronald W. Masulisa , Peter Kien Phama,∗, Jason Zeina , Alvin E. S. Angb,a a School of Banking and Finance, UNSW Business School, UNSW Sydney NSW 2052, Australia b Department of Economics and Finance, School of Business, The Hang Seng University of Hong Kong Shatin, Hong Kong September 20, 2021 Abstract We document a novel strategic motive for family business groups to utilize their internal capital markets (ICMs) during financial crises. We find that crisis-period group ICM activity is targeted toward exerting product market dominance over standalone rivals. Groups make significant post-crisis gains in market share that are concentrated among affiliates (and industry segments within affiliates) operating in highly competitive product markets, where capturing such gains is difficult in normal times. These patterns are observed only in emerging markets, suggesting that ICMs enable groups to exploit crises to realize long-term competitive advantages only when rivals face chronic financing frictions. Keywords: Family Business Groups, Product-market competition, Financial crisis JEL classification: G01, G31, G32 ∗ Corresponding author. Address: Room 307, UNSW Business School, UNSW Sydney; Phone: +61 2 9385 5875; Fax: +61 2 9385 6347; Email: peter.pham@unsw.edu.au Email addresses: ron.masulis@unsw.edu.au (Ronald W. Masulis), j.zein@unsw.edu.au (Jason Zein), alvinang@hsu.edu.hk (Alvin E. S. Ang)
“We believe that crisis is our opportunity for our future growth.” - Hyun-Suk Kim, CEO and President, Samsung Electronics.1 1. Introduction Around the world, a substantial fraction of publicly listed firms are members of business groups. These group structures arise when two or more firms become linked together via common ownership ties to a single controlling shareholder, often a wealthy family (Masulis et al., 2011). The extant literature has voiced numerous concerns about the corporate governance impact of business groups, arguing that families can use these structures to consolidate control and entrench themselves against outside shareholder intervention and government reforms.2 Yet, surprisingly limited empirical evidence is available regarding the link between business group prevalence and aggregate levels of competition in an economy. Morck et al. (2005) warn that business groups can create “economic entrenchment” by strategically deploying resources to perpetuate their dominance over other firms. Echoing this concern, Almeida and Wolfenzon (2006) posit a theory whereby such entrenchment can prevent the efficient allocation of capital to valuable standalone projects, while Boutin et al. (2013) provide evidence that groups’ economic power limits competition by preventing entry of new firms. While these studies highlight the broad and potentially serious consequences of business groups’ economic impacts, the conditions under which groups are able to expand their market power and strengthen their competitive positions over time are not well documented. Under the assumption that investor protection is continually improving over time (see Spamann, 2009), the incentives for families to control large business groups should dissipate as their ability to consume private benefits becomes more constrained (La Porta et al., 1999; Almeida and Wolfenzon, 2006). Yet, in many markets around the world, family groups continue to 1 “A rare look inside Samsung’s secretive ideas lab”, CNN.com, September 17, 2019. 2 See La Porta et al. (1999), Morck et al. (2005), Bertrand et al. (2008), Djankov et al. (2008), among others. 2
expand, with no end to their dominance in sight.3 For example, in South Korea from 2002 to 2012, a period which includes the Global Financial Crisis (GFC), the total annual sales as a percentage of GDP of the 10 largest business groups increased from 53% to 80%, with two thirds of the gain occurring after this crisis (Kwon, 2012). The above observations motivate our study’s research question: Do business groups exploit a crisis to expand their economic power? Thus far, the empirical evidence on this question in the business group literature is lacking. This is surprising given that internal financing benefits of group affiliation have long been associated with what the product market competition literature terms a “deep pockets” advantage, that is, the ability of financially strong firms to establish market dominance over financially constrained rivals (Maksimovic and Phillips, 2002; Faure-Grimaud and Inderst, 2005; Cestone and Fumagalli, 2005; Boutin et al., 2013).4 We argue that a business group’s ICM should provide its affiliates with a clear strategic advantage during crisis periods, allowing them to capture market share from standalone rivals with limited access to external capital. This echoes the observation in Phillips (1995) who states that “the deep purse has value when the capital markets are closed to the firm; otherwise, a firm can borrow when faced with ‘predatory’ behavior by rivals”. While past studies have analyzed the impact of business group membership on corporate investment levels during financial crises, the longer term strategic product market outcomes of such activity remain unclear. On the one hand, the evidence from Lins et al. (2013) suggests that the main strategic purpose of crisis period ICM re-allocations is to ensure the survival of the group’s affiliates rather than to expand its market power. In particular, using a multi-country sample of business groups, they find that in response to liquidity shocks precipitated by the 2008-2009 Global Financial Crisis, family groups cut investments (by more than non-family groups) to ensure the survival of their hard hit members and thereby 3 Masulis et al. (2015) show that there is a net increase in both the number of business groups and the number of listed group affiliates over the 2002-2007 period. They also show that the fraction of total market capitalization attributable to the largest business groups is also rising over the same period in some emerging markets. 4 See also Telser, 1966; Benoit, 1984; Brander and Lewis, 1988; Bolton and Scharfstein, 1990 for theoretical models on this theme. 3
preserve the family’s private benefits of control.5 Such actions are generally inconsistent with the goal of increasing market power. On the other hand, single-country evidence from two small economies, Korea (Almeida et al., 2015) and Chile (Buchuk et al., 2019), document that groups firms cut their invest- ment by less than similar non-group firms, especially for group affiliates with high growth opportunities (high Tobins Q). However, these studies do not consider the longer term strategic objectives that might be behind such investment behavior. Relatively greater crisis period investment by group firms is not necessarily motivated by the goal of dominating competitors.6 For example, while groups might utilize their ICMs during crises to help high Q affiliates meet their pre-existing investment plans, this is not a strategic response to an altered external financing environment. Rather, the decision to reallocate internal capital to these firms is simply a financially efficient response. From a strategic perspective however, this can yield little benefit to their competitive positions because a group’s high Q affiliates may already enjoy market dominance. Indeed, this may be the primary reason for their high Q (see Lindenberg and Ross (1981)). Our analysis recognizes that business groups do not operate in a competitive vacuum, but rather strategically interact with their rivals. We argue that such considerations can also explain how a business group utilizes its ICM during a crisis.7 Analyzing this aspect of a group’s behavior is important because it allows us to uncover a new explanation for the continued dominance of family business groups, namely exploiting crises as an opportunity to achieve product market dominance over their standalone rivals. Our analysis also recognizes that a group’s success in using their deep pockets during crises to establish longer term product market dominance depends critically on how effectively their competitors can counter these actions. This is likely to vary across countries depending 5 Massa et al. (2021) further argue that, during bad times, groups may transfer assets across affiliates in order to ensure the survival of their “central firms”. They show that such firms have lower risk and therefore lower expected returns. 6 Even in a general sense, there is no systematic and direct evidence linking investment levels to longer term changes in within-industry competitive positions 7 In a general context Fresard and Valta (2016) provide theory and evidence showing that firms’ investment decisions are strategically influenced by their competitive positions vis-à-vis their rivals 4
on the external financing environment. In developed capital markets, standalone firms are able to effectively respond to a group’s competitive threats by quickly regaining access to external finance when capital markets return to normal.8 If groups anticipate this response, then their incentive to use their deep pockets for competitive purposes can be seriously diminished. In contrast, for groups in emerging markets, any competitive gains they achieve during a crisis period can more easily be defended and built upon in the long run. This is because standalone firms will find it difficult to claw back product market share losses due to a chronic lack of access to external finance that persists even after capital market conditions return to normal. Using the 2008-2009 Global Financial Crisis (GFC) as our setting, we analyze the strategic responses to the crisis with a comprehensive sample of family business groups from 45 countries around the world. Given our coverage of both developed and emerging capital markets, the GFC is an ideal external financial shock as it effectively closes the external capital markets to virtually all firms around the world (Campello et al., 2010). Such conditions amplify a group’s deep-pockets advantage and allows us to more clearly observe its effects on product market competition. It is significantly more difficult to identify these same effects in normal times because the competitive benefits of deep pockets are likely to occur at a more incremental pace (or perhaps not at all), and thus are difficult to disentangle from other factors that could explain group dominance, such as unobservable controlling family skill levels and favorable government policies. Our baseline analysis relies on a difference-in-differences matching estimator methodology to compare the pre-to-post crisis changes in market share experienced by group firms to a matched set of control firms. The use of a matching estimator is critical in our setting because a common issue when studying business group affiliates is that they tend to be substantially larger than most firms in the local economy. Using OLS regression models in such a setting 8 It is also possible that in developed capital markets standalone firms face less severe declines in their access to external funds in financial crises periods, because they can access many alternative pools of capital, such as lines of credit and private placements with institutional investors. Also, competition regulators in developed markets tend to be more powerful and are more likely to limit the long term gains group firms can retain. 5
can lead to a violation of the common support assumption due to a lack of covariate overlap between treated and control firms. Our matching estimator approach ensures that we compare group firms to standalone firms with similar scale and financial capacity by matching on covariates such as size, cash holdings, industry sector, and pre-crisis market share. Our cross country approach also allows us to exclude from the analysis, countries where our matching procedure is unable to close the size gap between group and non-group firms. This helps to ease concerns that group affiliation is correlated with other underlying firm characteristics that could also explain a group’s ability to gain market share in the aftermath of the crisis. Our results show a significant pre-to-post crisis increase in product market shares of family group firms relative to similar non-groups firms, but only in emerging markets. Here, we find that groups expand their relative market shares by approximately 0.54 percentage points more than matched control firms over a 3-year horizon starting at the onset of the crisis. This rises to 1.13 percentage points over a longer 5-year horizon, or about one fifth of the average pre-crisis market share in a typical emerging market. We detect no such differences in developed markets, consistent with the notion that groups have limited capacity to generate long-term strategic benefits from a transitory shock when capital markets are resilient and well-functioning. Both the results for emerging and developed markets are robust to a variety of alternative matching criteria, including those related to industry definitions. To ensure that the market share increase of group firms observed for emerging markets is specific to the crisis period, we also examine other years before and after the GFC as placebo crisis years, but they do not produce any significant difference-in-differences in market share. We next specifically examine the strategic dimension of groups’ crisis-period ICM activity. First, we analyze whether a group deploys its deep pockets advantage in industries where affiliated firms can realize the most valuable competitive gains. We argue that the crisis opens up a strategic window of opportunity for group affiliates to establish their ascendancy in highly competitive industries, where in normal times, making such product market inroads can be too costly. To test this argument, we split the sample into two groups based on the pre-crisis levels of industry concentration (to capture rivalry among existing players) and by 6
average firm age in an industry (to proxy for ease in new firm entry). Our analysis confirms that product market gains are largely attributable to group firms operating in industries with high pre-crisis competition (measured by low market concentration and high entry rates). Second, we exploit the fact that firms often operate across multiple industries to examine how market share gains differ within a group firm’s industry segments. When examining industry segments where the group firm is not an industry leader, we find that their relative market share gains are significantly more pronounced compared to other industry segments where the group firm has already achieved market dominance. Another important aspect of our analysis is linking the observed gains in product market shares to the level of ICM activity in a group. Prior studies have shown that group ICMs become more active during crises to bridge the temporary gap between desired investment and financing capacity facing some group members (Almeida et al., 2015; Santioni et al., 2019; Buchuk et al., 2019). Since our focus is on whether group ICM re-allocations occur for strategic competitive reasons, we test whether groups who experience greater gains in product market shares are associated with higher crisis-period ICM activity. We identify business groups with particularly active crisis-period ICMs using two approaches. The first is based on a new measure of the size of intra-group investments, or investment in affiliates (where reporting is mandated by international accounting standards). The second approach involves instances of intra-group purchases of equity blocks. We show that both measures are associated with significantly higher increases in market share (of up to two percentage points over a 5-year horizon), but only for emerging markets. We explore several possible channels through which family groups might use their ICMs during the crisis period to grow their market shares. First, they may do this organically through their investment programs. For example, groups may maintain or even expand their production capacity and product development, in anticipation of poaching additional customers away from financially weakened rivals. We find that, during the crisis, family group firms in emerging markets cut their capital expenditures (CAPEX) by less than other firms. However, unlike previous studies, we clearly show that this difference is concentrated 7
in young industries with high pre-crisis levels of competition – the types of industries where market share gains are difficult to achieve in normal times. When we examine developed markets, we find that the crisis-period CAPEX of group affiliates actually declines more than other firms, consistent with our earlier market share results. We next attempt to demonstrate a connection between the above documented CAPEX patterns and a group’s product market strategies. To do this, we take a similar approach to Mukherjee et al. (2017) and rely on a textual analysis of company press releases, media articles and exchange announcements (available from the RavenPack News Analytics database), to identify events that would indicate a firm’s growing product market presence. We then analyze pre-to-post crisis changes in several types of product market expansion events, such as new product releases, entry into new markets, and production increases, for group versus non-group firms. We find in the aftermath of the crisis that these expansion events are more likely to occur in group firms than other firms. Overall, while our results for 22 emerging markets in our international sample are consistent with Almeida et al. (2015), who document higher capital expenditures by Korean Chaebols during the Asian Financial Crisis, our findings go one step further by connecting such patterns to a group realizing its strategic objective of expanding its market share. These results also point to an important channel through which groups can achieve superior post-crisis performance as documented by Almeida et al. (2015) and which we also find for our much larger and more diverse sample of business groups outlined below. Second, market share gains can occur through inorganic expansions, which we examine by analyzing groups’ acquisition behavior. We again find that in emerging markets the likelihood of group affiliates engaging in M&A activity increases during the crisis period relative to that of non-group firms. However, the same increase in group firm M&A activity is not observed in developed markets. Third, a group’s deep pockets can allow affiliated firms to sustain their operations as their competitors fail (or become financially distressed) during the GFC, leaving product market gaps available to be filled by group affiliates. Using a Cox proportional hazard model, we find that following the crisis, listed group affiliates in 8
emerging markets are significantly less likely to fail following the crisis, in line with the results in Santioni et al. (2019) for Italian firms. However, we also show that it is not simply group firms’ superior survivability that solely explains their increases in market share Of course, capturing market shares is not necessarily wealth creating for all shareholders, since it can involve costly investments incurred for the purpose of empire building or private rent seeking or expansions that cannot be sustained. Family groups may also willingly incur such shareholder wealth decreasing actions (e.g. large advertising campaigns) if product market dominance provides them with added private benefits such as greater political power and family-brand visibility (Morck et al., 2005). Groups may also underestimate the speed at which rivals can respond or overestimate their own ability to retain crisis period gains in market share. To assess whether growth in market share around financial crises is profitable, we analyze the buy-and-hold stock returns of group firms from the onset of the crisis period for up to five years thereafter. We show that group firms in emerging markets significantly outperform other similar non-group firms in terms of stock returns, indicating that family groups gains in market share do not represent over-investment. These results also expand on prior studies that document positive short-term consequence of business groups investments during crises (Almeida et al., 2015; Lins et al., 2013). By analyzing stock returns over a 5-year horizon from the onset of the crisis, we are better able to gauge whether new investments made in the depths of the stock market cycle reap long-term financial gains for family group firms in the recovery period. Our findings suggest that by exploiting strategic opportunities presented by financial crises, groups create long-term economic benefits for all their shareholders. Our study builds on the literature that shows how business groups can insulate their affiliates from external capital market shocks (Almeida et al., 2015; Santioni et al., 2019; Buchuk et al., 2019). However, none of these studies documents the longer term product market consequences of such actions.9 Our findings demonstrate that strategic imperatives 9 In the related context of (single-firm) conglomerates, Gopalan and Xie (2011), Matvos and Seru (2014), Kuppuswamy and Villalonga (2016), and Matvos et al. (2018) also show that ICMs of US multi-division firms are more active during episodes of severe external capital market dislocation. 9
to enhance their competitive positions and expand their economic influence are additional group objectives pursued through their crisis-period ICM activity.10 Our results also highlight an important connection between the business group and industrial organization literatures that focuses on the interaction between financial decisions and product market competition, such as the theoretical work of Bolton and Scharfstein (1990) and Faure-Grimaud and Inderst (2005). Empirically, Fresard (2010) makes use of exogenous decreases in barriers to competition due to tariff cuts to show that a firm’s financial strength (proxied by their cash reserves) leads to a systematic increase in their product market share, at the expense of industry rivals. In a business group setting, Kim (2016) analyzes competition among Korean business groups in the aftermath of the Asian Financial Crisis. She shows that groups in a position of relative financial strength achieve higher post-crisis sales growth rates compared to other groups. However, Kim’s study does not provide a comparison between group firms and standalone firms to demonstrate the consequence of having access to an ICM. Using a sample of French firms, Boutin et al. (2013) show that groups with deep-pockets negatively affect industry entry rates in normal times. Complementing these studies, our study provides cross-country evidence that makes use of a sudden and unanticipated global shock to external financing as a means to distinguish a group’s deep pockets benefits from other possible explanations (such as outstanding family talent). Our unique data and setting allow us to clearly document the considerable variations in the extent to which groups can capture these benefits – both across country-level long-term external financing conditions and across industry-specific competition environments. More broadly, our study contributes to the literature on the segmentation of world equity markets. Despite decades of strong domestic economic growth and financial globalization reforms, countries classified as emerging markets continue to lag behind developed markets in many respects including equity trading activity, capital raising, financial intermediaries and 10 Similar crisis-period benefits are documented in a private-equity setting by Bernstein et al. (2019). They show that portfolio firms of private equity funds with deep pockets (derived from the fund’s pre-committed, but untapped limited partner capital), are able to maintain investment and increase market share during the post-GFC period. 10
institutional investor development (Bekaert et al., 2011; Carrieri et al., 2013; Hanselaar et al., 2019), and they remain generally under-represented in international investors’ portfolios (Bekaert and Harvey, 2017). As argued by Almeida and Wolfenzon (2006), business groups may contribute to perpetuating these persistent differences across countries, as their ICMs can actually substitute for the role of financial intermediary lending within groups, thereby weakening the overall demand for the services of financial intermediaries and reducing external capital available to standalone firms. Our empirical results are consistent with this hypothesis by showing that family business groups in emerging markets are able to exploit financial crises to raise competitive pressures on their standalone rivals and to significantly strengthen their economic power through their use of deep pocket internal financing. 2. Data and sample construction 2.1. Business group sample Our empirical analysis requires the identification of business group affiliated firms from around the world in 2007, the year immediately preceding the onset of the GFC. We rely on the business group data assembled by Masulis et al. (2011) covering 45 countries as of 2002, and expanded to 2007 by Masulis et al. (2020). A key advantage of this dataset is its broad coverage of many developed and emerging market countries across five continents. This is achieved through a comprehensive procedure that combines standard ownership databases (Bureau van Dijk Osiris, Factset Lionshares, Thomson Reuters Global Ownership), hand- collected firm ownership data (from LexisNexis, Factiva, Bloomberg, Dun and Bradstreet’s Who Owns Whom, stock exchanges and securities regulators), and major transactions data (IPOs, M&A, etc.).11 It is important to recognize that level of ultimate ownership identification we obtain in our study cannot be achieved by relying solely on standard commercially available ownership databases. For example, Lins et al. (2013) rely on ownership information from Bureau van Dijk’s (BvD) suite of products (Osiris, Orbis, Amadeus, etc.), but we find that these databases 11 See Masulis et al. (2011) and Masulis et al. (2020) for more detailed descriptions of these data sources. 11
provide only a partial picture of actual business group ownership linkages. Out of the universe of all listed and delisted firms covered by the BvD databases in 2007, only about three quarters have any ownership data reported. Among them, only about 21 percent have ultimate owner information, as the databases only consider ownership chains connected by shareholdings of at least 25 percent and they do not consistently aggregate related blockholdings.12 Following Masulis et al. (2011), a business group is defined as two or more publicly-listed firms controlled by the same ultimate controlling shareholder. The control chain linking each firm to the ultimate controlling shareholder is established based on the largest ownership stake that is equivalent to having at least 20 percent of the voting rights in the firm (or 10 percent if the shareholder has some operating control as a founder, CEO, or board chair). Masulis et al. (2020) expand the Masulis et al. (2011) dataset by tracking how each group evolves over the 2003-2007 period. In summary, they use firms’ major transactions data to capture new groups formed and existing group expansions through IPOs (or spin-offs) of group affiliates and through partial acquisitions of new firms, as well as cases of groups divesting (liquidating) existing member firms. Masulis et al. (2020) then cross-check the snapshot of the business groups we identify against data from the ownership databases, Orbis, Worldscope, Global Ownership, and Lionshares to locate other missing group affiliated firms. We exclude financial firms from our analysis (with Standard Industry Classification (SIC) codes 6000-6999), given their unique status during the 2008-2009 financial crisis. In any sample of international firms, significant financial data and reporting anomalies can exist. We drop firms having negative cash holdings, negative total assets, negative book value of debt, negative common equity, cash-to-asset ratios exceeding one, and total assets ranked in the lowest 5th percentile in each country. After applying the above sample selection criteria, we obtain a sample of 19,803 listed firms from 45 countries as of 2007, of which 2,882 firms are 12 For example, in the Osiris database, the majority of listed firms in the Samsung business group cannot be ultimately traced to the Lee Kun-hee family as control is achieved through various fragmented blocks of less than 25 percent held directly or through affiliated firms. A comparison of our business group sample to that in Lins et al. (2013) illustrates the severity of this problem. When we restrict our sample to the same 35 country sample used in this prior study, our procedure leads to more than double the number of group affiliated firms. 12
affiliated with family business groups and 1,364 firms are affiliated with non-family groups (controlled by governments, financial institutions or widely held corporations). The remainder of the sample consists of 15,557 standalone firms, where their ultimate owner information can be ascertained to confirm that they are strictly unaffiliated with any type of business group, although they can be owned by a family. Table 1 provides a country-level breakdown of the sample. 2.2. Classification of capital market development We argue that capital market development reflects underlying cross-country differences in financing frictions. This link is well established in the literature, as prior studies have shown that market development is positively correlated with inbound portfolio investment flows (Chan et al., 2005), market trading and valuation (Bekaert et al., 2011; Carrieri et al., 2013), and development of legal institutions Djankov et al. (2008), all of which affect a firm’s access to external financing. Specifically, we use the MSCI index classification system to classify our sample countries into developed and emerging markets. The first cohort includes 23 “Developed Markets” that MSCI includes in the MSCI World Index. The second cohort comprises the other 22 sample countries that MSCI designates as “Emerging Markets” and “Frontier Markets” as of 2007 (referred to in our study as “Emerging Markets”). The list of emerging market countries is presented in Table 1. There are several important advantages of using the MSCI classification. First, it incorporates a wide range of criteria to capture the level of development in each national market, including (1) sustainability of economic development, (2) size and liquidity of listed firms, and (3) market accessibility to international investors.13 The application of these criteria is also vetted by the international investment community as MSCI seeks detailed feedback from institutional investors on its index decisions. Second, the classification is 13 This is arguably a more sophisticated approach than using only the relative size of a country’s stock market. For example, if our sample countries are grouped according to aggregate stock market capitalization scaled by GDP (with data from Djankov et al. (2008)), then Germany and Italy are below the median (and would be designated “Emerging Markets”). If they are grouped according to the number of listed firms per capita, then France, Germany, and the Netherlands are below the median. 13
widely adopted by international portfolio investors (with around $14.5 trillion of institutional funds benchmarked against MSCI indices as at 2020), resulting in significant differences in foreign fund flows into each of the two market classes (Ferreira and Matos, 2008; Burnham et al., 2018). Such foreign investments can directly impact a firm’s financing capacity and indirectly impact it through the role that foreign institutions play in improving corporate governance (Aggarwal et al., 2011). Third, there appears to be a clear and persistent divide between MSCI’s “Developed Markets” versus other markets. In the past three decades, there have been only 4 re-classifications affecting the developed markets list, involving Portugal (promoted in 1997), Greece (promoted in 2001 and demoted in 2013), and Israel (promoted in 2010).14 Based on the above classification, business group importance clearly differs across countries by market development status. Table 1 shows the breakdown of (family and non-family) group firms and standalone firms by country. Consistent with the conjecture that business groups thrive in an environment with high external financing barriers, we find that family (non-family) group firms on average account for 33% (10%) of sample firms in emerging markets, compared to 14% (6%) in developed markets. [Insert Table 1 here] An important element of our hypothesis is the relative weaknesses of emerging capital markets in supporting firms’ external financing. This point has been well established in the literature. For example, Henderson et al. (2006) show that both debt and equity issues are relatively limited outside the most developed capital markets. Doidge et al. (2013) find that domestic IPO activity intensifies over time as a country’s financial markets becomes more developed (due to financial globalization). Hanselaar et al. (2019) provide time-series evidence that equity issuance volume in emerging markets is both less frequent and less sensitive to market-wide liquidity improvements. 14 In fact, the emerging/developed markets divide is consistently recognized by all major index providers. For example, as of 2007, the MSCI list of emerging markets is mirrored in both the S&P/IFC and the FTSE Russell classifications. 14
Using our dataset, we also confirm that there is a clear divide between emerging and developed markets as reflected in firms’ general ability to raise external finance. In Appendix Table A1, we provide some preliminary tests using seasoned equity offering (SEO) and corporate investment (CAPEX) data to re-confirm this difference. In Column 1, we report that emerging market firms are significantly less likely to engage in SEOs than their developed market counterparts. This difference is equivalent to about 18% of the average rate of SEOs across all firms, and persists in the sub-sample of standalone firms (Column 2), which do not have access to an internal capital market. In Columns 3 and 4, the sensitivity of firm-level investment (measured by CAPEX/Assets) to internally generated cash flows (measured by net profits plus depreciation, scaled by total assets) is separately estimated for emerging market and developed market firms. This sensitivity captures the extent to which a firm faces external financing constraints, and our estimates confirm that this sensitivity is significantly higher in emerging markets, especially for standalone firms. 2.3. The 2008-2009 Global Financial Crisis Several features of the GFC make it an attractive external financing shock from which the impacts of a group’s deep pockets advantage can be inferred. First, capital markets in this period experience severe disruptions across much of the globe, with the crisis transmitted across both emerging and developed markets (Bekaert et al., 2014). Second, the onset of the crisis is both sudden and unanticipated, so it is unlikely that firms could preemptively make changes to their ownership structure in anticipation of the GFC. Third, unlike the 1997 Asian Financial Crisis, where business groups are often cited as a key trigger (Chang, 2006), in the GFC, business groups are not implicated as a potential cause. In contrast, the source of the GFC was centered in the United States, where family groups are not a dominant organizational form. Fourth, the crisis itself does not appear to be triggered by a large drop in corporate investment. Even the demand for loans does not fall by a substantial amount (Chang et al., 2019). In fact, practitioners, regulators and academics generally agree that overexposure of banks to subprime mortgage defaults is the primary trigger for the GFC, rather than it being the result of excessive corporate investment or debt financing. 15
The immediate consequence of the onset of the crisis is a severe contraction in credit availability. Ivashina and Scharfstein (2010) show that US banks, particularly ones with diminished deposit bases and larger outstanding credit lines, severely curtail their supply of new loans, beginning in late 2008. The equity issuance market is also adversely affected, with the aggregate seasoned equity offering (SEO) proceeds of global non-financial firms contracting from US$320 billion in 2007 to US$241 billion in 2008 (figures calculated using SDC Platinum data). Overall, this disruption to the supply of external financing at the country level leads to dramatic reductions in corporate investment, especially for financially constrained firms (Campello et al., 2010). In Figure 1, we show the global nature of the GFC. Across different regions around the globe, we observe a consistent decline of about 50 percent in aggregate stock market value, as approximated by regional MSCI indices. The plot also shows sharp declines during the second half of 2008 across all major regional stock indices, highlighting the unexpected nature of the crisis.15 This sharp fall coincides with the collapse of Lehman Brothers in September 2008, an event generally regarded as the beginning of a full-blown financial crisis. Given that we work with international firm data on an annual basis, we define the pre- and post-GFC periods for each firm using its financial year-end date. If a firm closes its books in the first two quarters of 2008, we define its last pre-GFC financial year as its 2008 financial year-end. If instead, a firm’s financial year-end is in the last two quarters of the year, then we define its last pre-GFC financial year to be 2007. [Insert Figure 1 here] 3. Group affiliation and crisis-induced changes in product market outcomes 3.1. Empirical predictions Our main analysis focuses on the long-term product market outcomes of family business groups following the 2008-2009 financial crisis. Our predictions are guided by the well- 15 Bekaert et al. (2014) show that firms in emerging markets are more widely affected by crisis contagion than firms in developed markets. 16
established “deep pockets” argument: a firm’s relative financial strength allows it to establish product market dominance over its rivals by being able to sustain losses (or maintain investment) without becoming insolvent (see Telser, 1966; Benoit, 1984; Brander and Lewis, 1988; Bolton and Scharfstein, 1990; Fresard, 2010).16 In contrast, financially weak firms (typically those that are highly leveraged) do not have this staying power, leaving them vulnerable to competitive pressure in financial crises.17 Our setting utilizes the GFC as a shock that substantially increases a group’s deep-pockets advantage relative to its rivals. Bringing the literature on the competitive effects of deep pockets to a cross-country setting, we argue that the incentives to use ICMs to capture product market share are likely to vary with different levels of external capital market development. A key consideration in extant theories of firms’ product market decisions is the ex-ante ability of rivals to effectively respond to such competitive threats. If a group firm observes that its rivals have sufficient financial capacity to weather a short-term loss of market share, then using their deep pockets to attempt to obtain market share gains may not be optimal. In line with this view, Fresard (2010) suggests that a rival’s cash holdings can deter competitive threats because of what he terms a “second strike” capability, that is, the rival can credibly signal that it has the financial capacity to retaliate. We predict that the capacity of standalone firms to effectively respond to competitive pressure from business groups is seriously restricted in emerging markets. Such restrictions are not confined to the crisis period, but also tend to persist in normal times. Thus, in equilibrium, the incentives for groups to use ICMs as tool to capture market share from standalone rivals should be much stronger in emerging markets, as groups know that chronic external financing constraints impair a standalone firm’s ability to effectively respond in these markets. This conjecture is supported by several studies that document significant constraints on access to external debt and equity capital in countries with weak financial development (see for example Henderson et al. (2006)). Hanselaar et al. (2019) further show 16 Several other studies document that conglomerate ICMs also provide a similar beneficial effect for their divisions (see Faure-Grimaud and Inderst (2005), Maksimovic and Phillips (2002)). 17 Opler and Tittman (1994) show that highly leveraged firms lose market share during industry downturns. 17
that even when aggregate market liquidity improves, firms in emerging markets continue to find it difficult raise external capital. Product market expansion may not be the only means through which groups can exploit their financing advantages during a crisis. Another hypothesis posited by past studies such as Lins et al. (2013); Massa et al. (2021) is that family-controlled groups have strong immediate survival concerns (to preserve families’ long-term private benefits of control). Hence, family-controlled groups may divert resources from growing affiliates to rescue other affiliates experiencing operating problems. This implies that a group may weigh up the benefits of using its ICM to strengthen their competitive positions against the benefits obtained from using the ICM to prop up its troubled affiliates. Our empirical analysis will assess the relative importance of these two alternative hypotheses in different external financing environments. 3.2. Constructing product market outcome changes following the GFC We examine product market outcome changes from the financial year ending immediately before the crisis (denoted Year -1 or the pre-crisis year) to up to five years after. Our proxy for product market outcomes is market share. To construct each firm’s market share, we sort our sample firms into industries according to the first two digits of their primary SIC codes. In an international setting, this is a complex task given that individual firms can operate in multiple industries and that their primary industry classification can change as their main activity switches over our sample period. Another complication is that vendors of international firm data, including Worldscope, generally do not provide time-varying (historical) industry classification. If researchers are only able to capture the latest industry classification, they would incorrectly obtain market share figures for certain firms that have switched their primary industries. The analysis would be affected by a systematic bias since often there can be structural issues that drive these firms to shift their primary activities from one industry to another.18 18 For example, from 2007 (the year before the GFC) until 2013 (the last year of our observation window), about one fifth of our sample firms change their primary industries. 18
We implement a detailed set of procedures to address the above issues. We begin by determining a firm’s 2007 primary industry SIC code drawn from an historical version of the Worldscope database. This provides a snapshot of our sample firms’ industry classifications at the onset of the GFC. We then use the industry segment sales information in Worldscope to identify cases where a firm changes its primary SIC industry in later years of our sample period. For firms with no segment sales information, we assume that their primary industries are unchanged. Similar to many past studies such as Campello (2006), Fresard (2010) and Billett et al. (2017), we assess product market performance of listed firms through their sales gains relative to their listed industry rivals. Specifically, we compute relative market share by first determining the total sales generated within every country’s specific 2-digit SIC industry and then calculating the percentage share of total industry sales attributable to each firm. This measure does not give us the precise market share of each firm given that it is not possible to capture all firms (including private firms) in each industry on a global scale. However, given our study’s hypothesis is about the competitive effect of a family business group’s “deep pockets”, the main difference that we seek to document is between a listed group firm and its listed rivals that are not group affiliated. For this objective, our market share measure still allows us to make such a comparison in an unbiased manner. Of course, there can still be excessive noise in this measure when an industry is sparsely populated by listed firms. Thus, similar to a criterion used in Fresard (2010) and Billett et al. (2017), we apply a minimum number of firms rule that only computes market shares for industries with at least five listed firms in a given country-year. We compute the market share changes for each firm’s primary industry from the pre-crisis year (Year -1) to three years later (Year +2) and to five years later (Year +4) where Year 0 is the crisis year. The first window captures the firm’s ability to withstand the immediate impact of the crisis and the second reflects its more long-term market share gains/losses. For firms that fail during one or both of these two measurement periods, we assign them a final 19
market share value of zero.19 This adjustment reflects the assumption that a failed firm is no longer operational and effectively loses all of its market share.20 Table 2 describes the time-series variations in a firm’s market share measure from 2003 to 2013. An important point to note is that there is a mechanical relationship between market shares and the number of listed firms in an industry. For emerging markets, the number of firms tends to increase during the sample period (see Columns 1 and 2). This means that for each firm in an emerging market, the market share measure is more likely to decline over time than it is to increase. In contrast, market shares in developed markets are relatively stable because the number of listed firms in each market does not systematically increase or decrease. 3.3. Matching methodology Accounting for unobserved differences between firms with and without family group affiliation is a significant challenge. For this reason, we may not be able to fully establish the causal effect of a family group’s deep pockets on its product market positions. However, if the 2008-2009 crisis can be assumed to be sudden and largely unanticipated, such that the industry structures of groups and strategies of firms across an economy do not adjust in anticipation of the crisis, then the analysis can still provide persuasive evidence on whether family-group firms are on average able to exploit a severe capital market shock to capture added market shares relative to other firms. This is important because of the somewhat mixed evidence in previous studies in relation to groups’ investment behavior during financial crises. For instance,Almeida et al. (2015) find that Korean Chaebols cut investments less 19 We define a failed firm as one that is delisted during the measurement period (either 3 or 5 years after Year -1) and satisfies one of the following conditions just before the delisting: (1) a final market capitalization value less than US$0.5 million, (2) a stock return over the period of less than -90%, and (3) zero reported sales. 20 This is not a material assumption. In Panel D of Appendix Table A4, we exclude firms that fail after the crisis (4.8% of the pre-crisis firm sample) from the analysis and show that the baseline results remain robust. We note that bankruptcy is only one of many possible ways that market share can change. In subsequent analysis, we empirically examine several channels for increasing market share: through organic growth (such as investments in product development and distribution channels and cutting prices (or lowering margins) to drive out rivals) and growth through acquisitions. 20
aggressively than non-Chaebol firms during the Asian Financial Crisis, whereas Lins et al. (2013) find that family group firms reduce investment more aggressively during the 2008-2009 crisis compared to non-family group firms using a multi-country sample. Our analysis has two important advantages over the empirical settings in these other studies. Because Almeida et al. (2015) focuses on a single emerging market, South Korea, where Chaebols dominate, it might be difficult to find appropriate counterfactual firms. For example, one key difference between group and non-group firms is their size. Evaluating distributional differences in firm size using the Kolmogorov-Smirnov test, Almeida et al. (2015) show that Chaebol firms are indeed much larger than non-Chaebol firms. The difference persists even after applying a matching procedure. Lins et al. (2013) compare the investment response of family group firms to non-family group firms using a relatively small cross-country sample of business groups and find that hard-hit family groups cut investments by more than non-family groups. However, their comparisons do not employ any matching methodologies. To address the covariate imbalance problem described above, we rely on a difference-in- differences matching estimator (DID-ME) developed by Abadie and Imbens (2006, 2011) to select appropriate matched firms and then use them as a benchmark to estimate the effect of the crisis on the market share changes of family group firms. This matching estimator is particularly attractive because it allows us to match on both categorical variables (such as country and industry) and on continuous variables that might predict product market success. Similar to Almeida et al. (2015), we argue that this is a more reliable approach to identifying crisis-induced deviations across firms than using a standard linear regression analysis, which can mask the fact that there is inadequate covariate overlap between family-group firms and the matched comparison sample of firms. Specifically, we form treatment and matched control samples based on information at the end of the pre-crisis year (Year -1). For each family group (subject) firm, we select from among the other firms in the same country and the same industry sector, the nearest neighbor match based on the following list of continuous covariates used to capture observable 21
differences in the ability of a firm to compete in its product market.21 First, we control for firm size (Size), measured by the logarithm of $US total assets, and firm age (Age), measured by the logarithm of the number of years since listing. Large and old firms may be at a life cycle stage where market share growth is relatively stable. Second, certain factors such as cash flows from operations, asset liquidity, existing financial leverage and asset tangibility may influence firms access to external financing, so our covariates include the following measures: net profits plus depreciation (Operating P rof it), cash holdings (Cash Holdings), book value of debt (Leverage), and the value of property, plant and equipment (P P E), all scaled by book value of total assets. Cash is a particularly important matching covariate, because we seek to distinguish the group firm’s deep pockets advantage via access to an ICM, from simply just being a cash rich firm. Finally, to account for investment requirements, the covariates also include capital expenditures scaled by total assets (Capex) and Tobin’s Q (Q), calculated as the market value of total assets (market value of equity plus the book value of debt) scaled by book value of total assets. Finally, to pick control firms from the same industry cohort, we use the first digit of a group firm’s SIC industry code.22 This approach represents our default matching procedure. A close examination of our data confirms that significant differences exist between family group firms and other firms along many dimensions. While this may create bias in a linear regression setting, we alleviate such problems using our matching estimator. In Appendix Table A2, we compare the frequency distribution of each of the above covariates across the two sub-samples: family group firms and other firms in the same country and same 1-digit SIC industry. Using the Kolmogorov-Smirnov test, we show that the two sub-samples significantly differ along all dimensions, except Q in emerging markets and Capex in developed markets. Focusing on size differences, we further report in Appendix Table A3 that the total assets distribution of group firms is significantly different to that of the rest of the firms in 13 out 21 We find that either including or excluding non-family group firms from the control group does not affect any of our main findings. See Panel A of Appendix Table A7. 22 This choice is mainly driven by the small number of existing firms in some countries and by the fact that our primary market share measure is computed at the 2-digit SIC code level. In robustness analysis discussed later, we change the above industry matching condition to the 2-digit SIC code level. 22
of our 19 emerging-market countries and 13 out of 22 developed-market countries (some countries are not included if our matching procedure fails to find matches for any sample family-group firm).23 By matching on the above categorical and continuous covariates, we are able to reduce the covariate imbalance between family group firms and their matched firms, but not completely eliminate it. Specifically, in Appendix Table A2, we show that our matching procedure closes the gaps in all covariates, except Size and Age. When delving into the size imbalance for each country (see Appendix Table A3), we show that the gaps are eliminated for 15 out of the 26 countries mentioned above where family group firms are systematically larger. Yet, significant size differences continue to exist in the remaining 11 markets.24 The richness of our cross-country sample means that, in later robustness analysis, we are able to exclude the countries where our matching remains unable to completely eliminate firm size differences, while retaining a reasonably large cross-country sample of matched firms (see Appendix Table A4). 3.4. Baseline results Figure 2 presents a graphical summary of our study’s main results, using two separate graphs for emerging and developed capital markets. Each graph takes a snapshot of our sample firms as of the pre-crisis year (Year -1) and then measures the average market share of family group firms and that of their matched control firms, for up to three years before the crisis and up to five years after (including the crisis year, Year 0). It is important to emphasize that, in our timeline, Year -1 is the last year before the crisis while Year 0 is actually the first year in which a firm faces the potential impact of the crisis. This is because a large number of firms have their financial year ending in December, so for them, Year 0 ends in December 2008, which is after the crisis has struck. The graph for emerging capital markets (Graph A) shows that, even before the crisis, 23 The median firm size, US$ total assets, of family-group firms is three times larger than that of the other firms in emerging markets, and 2.7 times larger in developed markets. 24 In many of these markets, groups are relatively important such as Italy, Indonesia, Singapore, South Korea and Turkey. 23
You can also read